Researchers traced the evolution of human music preferences, exploring the journey from classical to jazz.

A team of researchers from the Max Planck Society's Institute for Dynamics and Self-Organization (Germany) selected musical pieces stored in MIDI format for their experiments. The notes in these pieces can be easily interpreted and quickly processed on a computer.

To investigate the autocorrelation function of musical sequences, it was necessary to analyze one-dimensional time series. Therefore, priority was given to files corresponding to monophonic instruments (which can play only one note at a time, such as brass or woodwind instruments), or, in the case of polyphony, to the most prominent parts, such as the first violin. Compositions written for piano or harpsichord were excluded.

To characterize changes in stochastic time series, the physicists chose the method of power spectral density analysis. This method reveals periodic components as more or less pronounced peaks that are easily identifiable. In total, they evaluated 553 time series of pitch, of which 99 were extracted from classical music scores and 454 from jazz solo improvisations.

During the study, distinct peaks, as indicators of rhythmic structures, were observed at periods of 0.25, 0.5, 1, 2, and 4, corresponding to sixteenth, eighth, quarter, half, and whole notes. The researchers found that the peaks seen in classical pieces were generally much stronger than those in jazz improvisations. This difference can be attributed to the fact that classical music is rhythmically more regular and performed without anacrusis (an incomplete measure preceding the first beat of the following measure). The most common time signatures in both musical genres were 3/4 and 4/4.

The autocorrelation function describes the similarity between two series as a function of the time difference between them, and the physicists determined that for pitch, it initially declines very slowly. This allows for easy prediction of the musical process, i.e., the time series. However, the progression has a limit after which predictability sharply breaks down, leading to more surprises in the time series.

Jazz improvisations typically exhibited a shorter duration of this transition from correlated to uncorrelated expectations. Differences were also found in the perceptions of composers. For example, in Bach's works, the transitional period ranged from five to twelve quarter notes, while in Mozart's compositions, it spanned from eight to 22. This suggests that one can expect more variety from the music of the composer of "Toccata and Fugue in D minor" than from the masterpieces of the creator of "Symphony No. 40" and "Requiem." The predictability of musical development is higher in Mozart's works.

The researchers described the results of their search for a quantitative measure of musical diversity in an article published in the journal Nature Communications.